Highly acclaimed Madsen study (2002) is the wrong way to determine whether MMR causes autism

The Madsen 2002 study in Denmark is the most widely cited paper used to "prove" that vaccines do NOT cause autism. It's seriously flawed.

Executive summary

The Madsen 2002 study is the most widely cited study proving vaccines don’t cause autism. For example, Professor Martin Kulldorff cites this study as proof that vaccines don’t cause autism.

But the study is so seriously flawed that it is amazing that anyone still takes this study seriously.

The flaws

1. If we want to see if the MMR vaccine causes autism, the best and simplest way to do that is simply to plot every autism case relative to the closest MMR vaccination date of that child. The date is the date when the parent or expert observer first noticed telltale ASD behaviors (and not when a clinical diagnosis is made which could be years later). We’d make a plots for each dose as shown below. If the vaccine is causing autism, we’d expect to see a plot like the red lines.

   

2. You can make the same plot for the MMR unvaccinated by plotting the same thing but relative to the time that the child would have normally received the MMR shot per the CDC schedule. This would very likely produce a pattern like the green line.

3. The authors didn’t do that! Their way is to compare autism rates per person-years for the people with the MMR vaccine vs. people without the MMR vaccine. This would only really make sense if we followed every single person for say 2 months after the time they were supposed to have gotten the MMR shot (or actually did get the shot) and compared the number of cases. They didn’t do that.

4. So the short story is that they could have made this crystal clear using the graphs above, but instead chose a method that is not sensitive to the intervention.

5. Note that method #1 can be done retrospectively or prospectively. The advantage of a prospective trial is that the parent can make weekly videos before the shot and after the shot to objectively assess any behavioral changes and refine the ASD onset date.

Other problems with the study

1. The most important adjustments they should make are the factors that are known to impact the rate of autism. Gender and race are the two biggest ones. They completely ignored race! They adjusted for the mother’s education, for example. How that can possibly impact the rate of autism in a child is beyond me. Why these factors are even included in their adjustment model is questionable. If they are going to do adjustments, please first adjust only for gender and race and let’s see what that does alone. The others shouldn’t affect the results at all, and if they do, we have to wonder about biologic plausibility of that.

2. Their underlying data is unreliable. they’ve been notified of this, but ignore it.

3. They never validated the data they used to see if it was valid by checking with the parents.

4. The overall numbers of cases of autism were tiny. Heck, in my survey of just 10,000 parents, I found over 400 cases of autism, more than in this 8 year study of 537K children. Why hasn’t the CDC done a survey in the US of parents?

5. I asked to see the source data for the study. The author blocked me.

6. More issues listed in my earlier article on Professor Hviid.

7. In Denmark, they currently give the MMR vaccine at ages 15 months and 4 years. The 2002 paper said “The national vaccination program recommends that children be vaccinated at 15 months of age and again at 12 years.” So I guess they must have revised their recommendations. It’s amusing how they do this without any data to justify the change.

Comments from others on this article

Mark Blaxill, Brian Hooker, Mark Blaxill and I all - independently and by different means found this study to be fraudulent. I found they must have moved a number of patients from one diagnosis group to another - and calculated the exact number of patients they moved.

---

Hi Steve - your analysis is very good and quite insightful.  I have also heard that the Danish registry under ascertains vaccination status as a significant portion of vaccine records do not make it to the registry.  Indeed, some of the unvaccinated in the Madsen et al. 2002 study could be vaccinated.  This biases the study towards the null hypothesis which is to NOT find an association between the vaccine and autism which was the intent of the study in the first place.

Also, regarding your phone message, I do not know of any child who regressed into autism within 24 hours where the regression occurred prior to vaccination.  I do know of several cases where the child regressed into autism within 24 hours after vaccination.  Most notably, Temple Ealey received two MMR jabs inadvertently at his 12-month well-baby checkup.  His twin sister Lucinda Ealey didn't receive the MMR vaccine as Temple got the jab intended for her.  According to their mom, Dr. Sheila Ealey, Temple woke up completely different the next day and was completely unresponsive to touch, language, etc.

Other critiques of the Madsen study

But, of course, the biggest thing as ever with these studies is the perverse and opaque study design. They never designed these things so you can see at once what you are looking at. Very good for journalists since they can’t begin to understand what’s been done but it all looks very technical and sciency and beyond them. Even fellow scientists are usually very lazy and prefer to defer to opinion rather that spend a little time looking at it for themselves.

Study overview:   Admitting that previous studies which disproved the link between MMR vaccine and autism were limited by design and responding to the World Health Organization’s call for further study, the authors sought a larger data sample to increase the statistical analysis capabilities. They conducted a retrospective study on children born in Denmark between 1991 and 1998, after the MMR vaccine was approved for use in Denmark in 1987. They compared the cohort of unvaccinated children to the nearly half of million children who received the vaccine and found an insignificant relative risk of autism and autism-spectrum disorders in the vaccinated population.

Study's Flaws:  The data presented by the researchers in their findings is inconsistent, calling into question their entire methodology.  Table 1 reports 269 vaccinated children with autistic disorder and 47 unvaccinated children with autistic disorder.  Table 2 reports 263 vaccinated children with autistic disorder and 53 unvaccinated children with autistic disorder.  Depending on which set of numbers is used changes the results to other slightly less or slightly more than a Relative Risk of one when the vaccinated cohort is compared to the unvaccinated. That their findings are based on inconsistent data should be enough for credible reporting on vaccinations to not cite the study.  Also the authors themselves admit that this type of study does not rule out that a subpopulation could be vulnerable to the MMR vaccine as presented in the 1998 Lancet study [Thompson’s study].

Conflicts of interest:  This study was funded by the National Immunization Program (NIP) at the U.S. Centers for Disease Control and Prevention (CDC) which had a vested financial interest in increased uptake of the MMR vaccine as it directly bought the vaccine from Merck and distributed it for reimbursement to the states’ public health departments.  Also, one of the co-authors of the study, Dr. Diana Schendel, was a CDC employee at the time of publication.  In addition, three of the co-authors (Dr. Mads Melbye, Mr. Jan Wohlfahrt and Mr. Anders Hviid) were employees of Staten Serum Institut, a for-profit company that manufactures and distributes vaccines in Denmark.

Other studies of the Denmark data that do show an association between MMR and autism

See the Goldman study.

A note about my perspective

I focus on autism because it's important and the data is crystal clear.

Once parents worldwide are red-pilled on this topic, they will be furious, demand accountability, and be open to hearing other truths, the biggest being that vaccine risk benefits are in aggregate negative; there may be a good vaccine, but there isn’t any data to prove that because they hide the data from the public. Nobody should be getting any shots until the record-level public health data regarding these shots is made public which I have called for many time (and which the medical community refuses to call for).

See also this tweet:

Summary

Whether vaccines cause autism is a critical issue to get right.

The medical community has had decades to get it right with a well-designed study and they have completely failed. They have simply never done a study using the simplest, most obvious, and most sensitive method to find whether there is a signal or not (described in step #1 above). This study would be trivial to do in VSD or in a few large pediatric practices. Why was it never done? And why don’t they make this data public so anyone can analyze it?

Instead, they embrace a methodology that is clearly inferior and a study that is clearly flawed and then declared the issue settled science.

It’s almost as if nobody wants to know the truth, isn’t it?

Now you know just how inept they are.

And it should cause them all great embarrassment that this is all being pointed out by an MIT electrical engineer.